Despite the organization of the massive influx of multimedia content being the central question in development of informational technologies for last several decades, significant progress in the development of technologies for semantic-based storage and retrieval of naturally occurring content is still missing. The dissertation is focused on this technical issue of contemporary information technologies. By analyzing current approaches to machine interpretation of visual communications and to the development of content recommendation systems, and on the basis of accumulated knowledge of graphic engineers and numerous corrective insights of the modern theory of statistics, te dissertation will develop a new methodology for optimizing visual communication through the construction of an interpretable joint distribution space of basic informational features in visual compositions represented through quantification of their universal stylistic determinants. Through the introductory chapters, the dissertation offers a discussion of the stakes and measures of success of modern informational technologies, and raises concerns about their congruence with established aspirations of the Information Age. Attention is paid to the burning problem of developing information technologies for creating, designing, archiving and, in general, accommodating the modern torrent of informational contents. The introductory chapter presents this issue through the history of the development of computer systems for marking, storing and retrieving content. The problems that arise with attempts to mark the content linguistically for the purpose of their organization and the possibility of targeted retrieval are highlighted. It is explained that these problems arise due to a misunderstanding of the nature of interpersonal messages and due to the hasty adoption of layman’s and unsubstantiated views on the role of forms and the nature of symbols through which communication is realized that form the basis of todays informational technologies. The paper recommends paying attention to the contributions of the graphic profession as a profession that through centuries of direct experience working on informational products and building technologies for mass communication of messages, has accumulated knowledge for understanding the nature of information and tools for handling it. The dissertation then offers an overview of the field of work, namely the field of computer aesthetics, where it presents an overview of the features of the form of essentialmessages for assessing their communication value as recognized and formalized within the graphic profession, and presents a historical overview of the development of ways of measuring them using computer technologies. Setting the original framework for a complete overview of the field of development of informational technologies with an emphasis on the development of algorithms in the field of computer aesthetics, the successes and limitations of the tested approaches to the problem are clearly presented. The research field of computational aesthetics is a crucial contributor to the development of mechanisms for filtering and/or generating value-laden informational content. This dissertation acknowledges a recognized escalating problem in the development of contemporary informational technologies and presents a practical solution for communicational quality management by employing an innovative approach to the computational aesthetic evaluation (CAE). After presenting the hypotheses and expected contributions, through the chapter on methodology the gradual course of research in the dissertation is clearly presented through the development of algorithms for measuring the communicative value of messages through their measures on two basic semantic dimensions - in the dissertation called dimensions of width and depth of discussion, bearing meaning related to common usage of those expressions. This research step consisted of developing a program for the careful selection of examples of art compositions that will form a representative sample of successful visual communications; selection and design of a computer vision model architecture that, after training, will recognize important semantic features via performing computations on the spectrophotometric values of the image record; and defining the implementation of the model training, cost function, learning rate and the criteria for successful completion of the training. After discussing the problem and attempted approaches to its alleviation, dissertation offers a novel expert solution by presenting an original research approach and its resulting opensourced model which outperforms its current state-of-the-art competition in semantic and stylistic classification, at the same time providing an idiomatic measure for objective aesthetic evaluation and demonstrating semantically rich and professionally recognized explanatory power which can serve as the solid basis for development of reliable and user friendly content retrieval, generative or auxiliary design applications. Presented model is resource- and privacy-wise upmostconservative. Its use evades all ethical, legal or security concerns that beset all currently prominent models. Its developmental and operational costs are practically nil. In conclusion, this study presents a novel approach to semantic feature detection using convolutional neural networks fine-tuned on stylistically conditioned images of artworks from WikiArt. The models achieved high F1 scores and provided robust results in distinguishing between abstract-concrete (breadth) and iconic-symbolic (depth) dimensions of artistic expression. The findings suggest that the proposed method can be used for universal semantic feature detection, aiding in the interpretation of visual communications across different, past and future cultural periods. Additionally, the models provide an idiomatic aesthetic evaluation criterion, allowing for the assessment of artworks’ success in visual communication based on their positioning within the semantic space. Further research could explore expanding the model’s applicability to include more diverse and historically representative samples of artwork, as well as investigating the potential applications of this method in other domains such as graphic or multimedia design.